- Email: [email protected]
COMPARISON OF MOLECULAR AND CONVENTIONAL STRATEGIES FOR FOLLOWUP OF SUPERFICIAL BLADDER CANCER USING DECISION ANALYSIS
0022-5347/00/1633-0752/0 THE JOURNAL OF UROLOGY® Copyright © 2000 by AMERICAN UROLOGICAL ASSOCIATION, INC.®
Vol. 163, 752–757, March 2000 Printed in U.S.A.
COMPARISON OF MOLECULAR AND CONVENTIONAL STRATEGIES FOR FOLLOWUP OF SUPERFICIAL BLADDER CANCER USING DECISION ANALYSIS ROBERT K. NAM, DONALD A. REDELMEIER, PHILIPPE E. SPIESS, HEATHER A. SAMPSON, YVES FRADET* AND MICHAEL A. S. JEWETT From the Division of Urology, Princess Margaret Hospital, University Health Network, Department of Medicine, Sunnybrook and Women’s College Health Sciences Centre, University of Toronto, Toronto, Ontario, and Division of Urology, Laval University, Laval, Quebec, Canada
ABSTRACT
Purpose: Patients with superficial bladder cancer require long-term surveillance for recurrence. We compared the cost of cystoscopy and cytology (standard care) to that of urinary markers (modified care) for patients with a history of superficial bladder cancer. Materials and Methods: We constructed a decision analysis model that compared the 2 strategies for a hypothetical followup interval of 3 years. Probabilities required for the decision tree were based on a cohort of 361 patients diagnosed with superficial bladder cancer from 1987 to 1997. Sensitivity analyses were used to determine whether test sensitivity and specificity would affect cost thresholds. Costs for each strategy were then applied to actual practice patterns. Results: The cost of modified care ranged from $158 to $228 for each followup visit when using a urinary marker with a sensitivity and specificity of 95% and 77%, respectively. The cost of standard care was $240 for each followup visit. Based on sensitivity analyses the probability of disease recurrence and urinary marker accuracy were important determinants of expected costs. Mean number of followup assessments for patients followed more than 3 years was 4.3, 2.2 and 1.5 for years 1, 2 and 3, respectively. Cumulative costs of modified care were lower than those of standard care. Conclusions: Urinary marker testing for followup of patients with superficial bladder cancer is less expensive than the standard method of cystoscopy and urinary cytology based on our model. Future studies will be required to consider other factors that could affect the cost advantage of urinary markers, including indirect costs, the psychosocial impact of testing and different surveillance frequencies. KEY WORDS: bladder neoplasms; tumor markers, biological; decision support techniques; costs and cost analysis
Bladder cancer is common, lethal and potentially curable.1– 4 Up to 80% of all superficial bladder cancers will recur and 11% to 28% will progress1– 4 and, thus, early detection and treatment of recurrence are essential.5 This rationale is based on premises that bladder cancer progression results in metastasis and death, and treating disease recurrence delays progression. The former premise has been well established as 78% of patients with progression of superficial bladder cancer eventually die of bladder cancer.1 The latter premise has yet to be directly established to our knowledge. Since there is no untreated natural history data to compare whether treating disease recurrence prevents clinical progression, we can only infer this premise indirectly. Holma¨ng et al showed an inverse relationship between the rate of disease progression, and tumor size and volume among patients treated for superficial bladder cancer.2 Thus, detecting disease recurrence earlier would result in smaller tumor size and volume and, subsequently, lower progression rates. Accordingly, patients undergo surveillance cystoscopy and urinary cytology every 3 months for 1 to 2 years, every 6 months for the subsequent years and every year thereafter.6, 7 Such
testing results in anxiety, discomfort and costs.7 Furthermore, cystoscopy is invasive and can cause urinary tract infection in 15% of patients, dysuria in 59%, urinary frequency in 41% and hematuria in 30%.8, 9 Urinary cytology is an ancillary method for detecting recurrence but is associated with significant inaccuracy due to specimen handling techniques,10 variable observer reliability,11 large test performance differences between tumor stages and grades, and variation induced by intravesical treatments.12, 13 Other strategies to improve surveillance have been proposed. Adjuvant intravesical therapy and re-resection of the initial tumor have been shown to reduce recurrence rates and allow less frequent followup.4, 7, 14, 15 Tailoring the frequency of surveillance cystoscopy according to the risk of recurrence also appears safe.5, 7 More recently urinary markers for disease recurrence have been developed and commercialized16 –20 which require use of immunoassay techniques to detect tumor cells.16, 18, 19, 21 Marker testing only requires voided urine specimens to produce objective results and has the potential to reduce health care costs as well as morbidity associated with invasive surveillance methods. However, urinary marker testing needs to be accurate to serve as a safe and cost-effective adjunct or as a substitute for followup cystoscopy. To understand the use of these markers better we compared the costs of urinary markers and followup cystoscopy to detect disease recurrence in a cohort of patients with
Accepted for publication October 15, 1999. Supported by a grant from Diagnocure, Inc., Sainte-Foy, Quebec, and the Surgical Scientist Program, Department of Surgery, University of Toronto, Canada. * Financial interest and/or other relationship with Diagnocure, Inc. 752
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
superficial bladder cancer through a cost minimization analysis using decision analytic techniques. MATERIALS AND METHODS
Historical cohort. We studied patients treated at the Toronto Hospital and by the Quebec Urologic Group22 between 1983 and 1997 who were disease-free at first followup after initial treatment for superficial transitional carcinoma of the bladder (stages Ta and T123). The primary end point of biopsy proved recurrence was analyzed using nonconstant hazard functions and Kaplan-Meier survival analysis from the time of the first disease-free assessment.24 An event was defined as recurrence. Patients were considered to be at risk from the date of first followup until that of recurrence or last followup. The effect of tumor stage and grade was examined by univariate and multivariate analyses using computer software. Decision model. We constructed a decision analysis model to estimate the costs of followup with cystoscopy and urinary cytology (standard care), and a urinary molecular marker (modified care) to diagnose recurrence (fig. 1). The outcome for the decision tree was disease recurrence. We used the societal perspective, assumed a time horizon of 3 years and applied no discount factor. Entry criteria for the decision tree consisted of disease-free status at first followup visit, pathological stage Ta or T1 transitional cell carcinoma of the bladder with no evidence of carcinoma in situ and a positive test result with the urinary marker at diagnosis (90% detection rate17). The base case analysis was based on a patient of any age with superficial bladder cancer of any grade and stage requiring surveillance for disease recurrence with followup cystoscopy and cytology or the urinary marker used at the Toronto Hospital (Immunocyt†) with a reported 95% sensitivity and 77% specificity.16 The base case describes the patient for whom the expected costs derived from the model will apply. Probabilities for each branch of the decision tree were based on the hazard rates for recurrence in our cohort, and the reported sensitivity and specificity of our urinary marker.16 Various hazard functions were tested and the func† Diagnocure Inc., Quebec, Canada
753
tion that yielded the highest correlation to Kaplan-Meier survival estimates was used. Baseline assumptions. Major assumptions for the decision tree included that the gold standard of cystoscopy and cytology is 100% sensitive and specific; some tumors will test marker negative and may continue to be negative despite recurrence which would lead to false-negative results; a positive result from the urinary marker requires confirmatory cystoscopy; undetected recurrent tumors will not always produce hematuria or symptoms before progression, specifically 20% will be assumed to progress silently; all patients with progression will require radical cystectomy; patients will undergo followup evaluation every 3 months for year 1, every 6 months for year 2 and once for year 3; and if a patient has recurrence during surveillance, the risk after treatment will be the same as the initial risk of recurrence. Standard care consisted of cystoscopy at a hospital ambulatory facility and urinary cytology at each patient visit. The costs comprised professional fees of the urologist and pathologist, personnel time for cystoscopy and pathology, and institutional overhead costs, excluding depreciation. The cost of communicating test results, such as transportation and parking costs to the patient, was not included. Modified care costs included the reported cost ranges of commercially available urinary markers (kit costs assumed the entire kit was used).16, 18 –20, 25 Laboratory staff costs were estimated by the manufacturer and included the per test charge, and all handling and transportation fees. Sensitivity analysis. We examined all key variables of the model over a wide range of values using sensitivity analysis. Plausible ranges of the urinary marker sensitivity and specificity were based on reported ranges of other commercially available urinary markers.16, 18 –20, 26 We determined the cost thresholds (point at which 2 strategies have equal expected costs) for a range of sensitivities and specificities of the urinary marker, which was performed within the covariate categories (disease stage and grade) that were found to be significant prognostic factors for tumor recurrence within our cohort. Because costs of standard and modified care vary between centers, we tested the sensitivity analysis over a range of costs for each management option.
FIG. 1. Decision tree for patients at each followup visit. UM, urinary marker
754
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
TABLE 1. Baseline characteristics among patients with superficial bladder cancer who were disease-free at first followup evaluation Disease Stage Ta T1 Totals
No. Histological Grade (%) 1
2
3
Totals
115 7 122 (34)
175 29 204 (56)
10 25 35 (10)
300 (83) 61 (17) 361
RESULTS
Likelihood for disease recurrence. A total of 361 patients were disease-free at first followup assessment after initial resection and satisfied inclusion into our historical cohort analysis (table 1). Mean patient age at diagnosis was 64.9 years (range 29.3 to 93.2) and mean followup was 17.6 months (range 0.1 to 119). Based on correlational analysis of the survival estimates of various hazard functions and Kaplan-Meier survival estimates, the log logistic hazard model showed the highest correlation (Pearson r ⫽ 0.987, p ⫽ 0.0001). Univariate analysis based on the log logistic hazard model showed tumor stage (p ⫽ 0.0036) (fig. 2) and grade (p ⫽ 0.007) to be significant predictors of recurrence-free survival. Multivariate analysis revealed that tumor stage (p ⫽ 0.05) was a more important predictor than grade (p ⫽ 0.09). Costs of care. At the time of analysis the cost of cystoscopy was $221, including $89 for the professional fee, and $132 for facility and equipment. Urinary cytology was $19 per test, including $7 for technical and $12 for professional fees. Total cost of radical cystectomy was $9,570, which included $1,900 for professional fees and $7,670 for institutional fees (hospital wards, blood bank, laboratory, radiological expenses and so forth). The cost of the urinary marker used was $6016 and others ranged from $12 to $125.20, 27 Baseline analysis and sensitivity analysis. Because tumor stage was a stronger predictor of disease recurrence than grade based on the survival model, the cost of modified care was calculated for each followup visit by tumor stage. For the base case the cost of modified care was highest at the 3-month followup visit ($228 for stage Ta, $237 for T1) and decreased for every subsequent visit (table 2). The expected cost of modified care was consistently lower than that of standard care, which was $240 for every followup visit. Based on sensitivity analysis the probabilities for recurrence and urinary marker sensitivity and specificity were
important determinants of the expected costs of modified care. The later the followup evaluation, the lower the sensitivity and specificity of the urinary marker required for modified care to have a cost savings over standard care (table 3). Using the base case 95% sensitivity and 77% specificity as references for 1-way sensitivity analysis, the minimum urinary sensitivity required for modified care to have a cost advantage was 70% at the 18-month followup and minimum specificity was 37% at the 36-month followup (table 3). To compare other costs of urinary markers and standard care, multiple 2-way sensitivity analyses for urinary marker sensitivity and specificity were performed using a fixed recurrence rate of 45.8%, the highest probability (and most conservative) for all followup evaluations. A larger range of urinary marker sensitivities and specificities were plausible at lower costs of the urinary marker ($20) and higher costs of standard care ($300) (fig. 3). As urinary marker costs increased and standard care costs decreased, the urinary marker sensitivities and specificities required for modified care to have a cost savings increased. To determine how the use of modified care would affect actual practice patterns we compared the cumulative costs of modified and standard care for our cohort. Mean number of followup evaluations (cystoscopy and cytology) for patients followed at least 3 years was 4.27 (range 2 to 6) for year 1, 2.20 (range 1 to 4) for year 2 and 1.48 (range 1 to 4) for year 3. This number was determined individually by examining the frequency of followup for the first 3 years. Using our base case urinary marker the cumulative cost of modified care was less than that of standard care after 6 months of followup (fig. 4). DISCUSSION
Urinary molecular markers may provide cost savings compared to standard cystoscopy and urinary cytology for superficial bladder cancer followup. The first assumption based on clinical perception is that to consider modified care urinary marker sensitivity must be close to 90% or better. The disutility of a false-positive or negative result was not included in our analysis, primarily because we are unaware of quantitative reports of the psychosocial impact of false-positive or negative test results with urinary markers or cystoscopy which makes it difficult to calculate the cost impact. Because these indirect costs were not included in our analysis, we did not consider costs to patients for time spent undergoing cytoscopy, including lost wages, transportation and so forth. Also, it is noteworthy
FIG. 2. Kaplan-Meier recurrence-free survival by stage of superficial bladder cancer
755
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP TABLE 2. Base case analysis of expected cost of modified care using the Immunocyt urinary marker Followup Mo.
Stage Ta Ca ($)
Stage T1 Ca ($)
3 228 6 203 9 185 12 173 18 165 24 158 36 158 Expected cost for standard care was $240 for all followup
237 207 188 175 166 159 159 visits.
that our model does not consider how the use of urinary markers might affect disease specific survival. We assumed that the risks of progression and death were equal with both strategies and to consider how this would affect our model, an analysis of quality adjusted life expectancy would be required, which is beyond the scope of our study. Given these limitations, we have provided a simple method for decision markers to evaluate whether urinary markers would have cost savings over standard care for an institution. The acceptable cost of a marker with known test performance
TABLE 3. One-way sensitivity analyses of urinary marker sensitivities and specificities for each followup visit Followup Mo. 3
6
9
12
18
24
36
% Stage Ta sensitivity 93 88 82 76 70 Not significant Not significant % Stage Ta specificity 69 54 46 41 38 36 36 % Stage T1 sensitivity 95 89 83 77 71 Not significant Not significant % Stage T1 specificity 75 56 47 42 39 37 37 Thresholds were defined as the minimal sensitivity and specificity for which the expected cost of modified care is less than that of standard care, and baseline was 95% sensitivity and 77% specificity.
FIG. 3. Comparison of cost of cystoscopy and cytology (SC, $100 to $300) to that of urinary marker (MC, $20 to $100) using multiple 2-way sensitivity analyses. Cost advantage of bladder cancer followup by urinary marker is represented by black areas according to urinary marker accuracy. Probability of disease recurrence is fixed at 46%.
756
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
FIG. 4. Cumulative cost comparison between standard care (SC) and modified care (MC) using costs at Toronto Hospital, urinary marker test performance of 95% sensitivity and 77% specificity, and actual practice patterns of followup frequency.
can be calculated using local costs of standard care (cystoscopy plus cytology, fig. 3). Alternatively, with known local costs of standard care and a urinary marker, the required minimum test performance for that marker can be calculated. The use of other commercially available urinary markers appears to be financially viable depending on institutional costs of standard and modified care. Other assumptions made for the decision tree have been biased in favor of standard care and several need to be examined. In our model no confirmatory cystoscopy or cytology is done for patients with a negative marker test. Thus, tumor recurrence might be missed which could subsequently progress. Nearly all undetected tumors will ultimately present with hematuria or other symptoms, and the overall progression rate of low risk stage Ta, T1, grade 1–2 tumors is exceedingly low.28, 29 Furthermore, in a study that compared surveillance cystoscopy and a urinary lipid marker for disease recurrence among patients with superficial bladder cancer for 6 months after initial treatment the 27% with a false-negative test result had no clinical or stage progression.30 We assumed a 20% progression rate for patients with a false-negative urinary marker result and that they were curable with radical cystectomy. Based on available published data the 20% rate is probably a significant overestimate.22 Although the risk of recurrence positively correlated to the number of previous recurrences,3 we did not stratify each pretest likelihood of recurrence by the number of recurrences, because a variety of patient and physician dependent factors made it difficult. Patients may have missed scheduled appointments due to illness or vacation, or physicians may be overly cautious and choose to do more frequent followup evaluations. To assign and rank each visit would require an iterative process of examining each patient pattern of followup that would be difficult in a large cohort. These factors are also likely to explain why the frequency of followup evaluations was higher than expected. Furthermore, we did not stratify patients based on other important prognostic factors, such as tumor grade,3 because of the limited sample size of our cohort. Thus, we used the strongest prognostic factor (tumor stage) found from our cohort for our decision model. It would be important for future studies with a larger sample size to consider other important prognostic factors.
Also, the probability of recurrence was the most important determinant of cost associated with modified care using urinary markers. In our study the use of urinary markers was applied to the entire cohort. However, the cost of urinary markers would be significantly reduced if applied to the 60% of patients at low risk of recurrence based on simple clinical findings, such as the number of tumors and results of the first followup cystoscopy.3 Furthermore, for patients with low grade papillary tumors cytology may not provide additional diagnostic information given its low sensitivity, and cystoscopy alone may be used which could affect the cost advantage of urinary markers.12 However, because the most important predictors of cost within the decision tree were the probability of disease recurrence, and urinary marker sensitivity and specificity, the reduction in cost of standard care for cystoscopy alone did not significantly affect expected costs. As well, the gold standard of cystoscopy and cytology could be used at different followup intervals. Olsen et al showed that lowering the frequency of followup cystoscopy in select patients may not affect disease progression rates.5 It would be important to consider these possibilities further in future studies. Finally, molecular techniques may even provide a superior method for early detection of recurrence. Although we used the combination of cystoscopy and cytology as the gold standard, microscopic lesions may still be missed. It is possible that a urinary marker test may detect the presence of occult bladder cancer recurrence when conventional methods do not. In such a case a true positive result with a urinary marker test may be classified incorrectly as false-positive if the gold standard cystoscopy is negative. Early reports on urinary marker followup of patients after initial resection have revealed that some patients with negative cystoscopy but a positive marker urinary test have bladder lesions with further followup.26 In our analysis this bias would favor the more conservative standard care method. CONCLUSIONS
From the payer perspective urinary marker testing for followup of patients with superficial bladder cancer can provide lower costs than the standard method of cystoscopy and urinary cytology. Reported sensitivities and specificities of various uri-
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
nary markers are sufficient to be cost saving per patient within 3 years of followup. The cost advantage of urinary markers over standard care remains over a wide range of service costs and increases with time of followup. The actual frequency of followup for patients on surveillance for superficial bladder cancer may be higher than what is recommended. In this setting urinary marker testing showed an even higher cumulative cost savings over standard care. This method is also more appealing to patients since only a voided urine specimen is required, and it avoids the morbidity associated with invasive testing. REFERENCES
1. Herr, H. W.: Natural history of superficial bladder tumors: 10- to 20-year follow-up of treated patients. World J Urol, 15: 84, 1997 2. Holma¨ng, S., Hedelin, H., Anderstro¨m, C. et al: The relationship among multiple recurrences, progression and prognosis of patients with stages TA and T1 transitional cell cancer of the bladder followed for at least 20 years. J Urol, 153: 1823, 1995 3. Kurth, K. H., Denis, L., Bouffioux, C. et al: Factors affecting recurrence and progression in superficial bladder tumours. Eur J Cancer, 31A: 1840, 1995 4. Pawinksi, A., Sylvester, R., Kurth, K. H. et al: A combined analysis of European Organization for Research and Treatment of Cancer, and Medical Research Council randomized clinical trials for the prophylactic treatment of stage TaT1 bladder cancer. J Urol, 156: 1934, 1996 5. Olsen, L. H. and Genster, H. G.: Prolonging follow-up intervals for non-invasive bladder tumors: a randomized controlled trial. Scand J Urol Nephrol, Suppl., 172: 33, 1995 6. Messing, E. M. and Catalona, W.: Urothelial tumors of the urinary tract. In: Campbells Urology. Edited by P. C. Walsh, A. B. Retik, E. D. Vaughan, Jr. et al. Philadelphia: W. B. Saunders, vol. 3, chapt. 77 pp. 2327–2410. 1998 7. Hall, R. R., Parmar, M. K., Richards, A. B. et al: Proposal for changes in cystoscopic follow up of patients with bladder cancer and adjuvant intravesical chemotherapy. BMJ, 308: 257, 1994 8. Westenfelder, M., Rosset, K. and Pelz, K.: Development of nosocomial and iatrogenic urinary tract infections (UTI) following urological interventions. A prospective clinical study. Scand J Urol Nephrol, Suppl., 104: 59, 1987 9. Denholm, S. W., Conn, I. G., Newsam, J. E. et al: Morbidity following cystoscopy: comparison of flexible and rigid techniques. Br J Urol, 66: 152, 1990 10. Leyh, H. and Mazeman, E.: Bard BTA test compared with voided urine cytology in the diagnosis of recurrent bladder cancer. Eur Urol, 32: 425, 1997 11. Sherman, A. B., Koss, L. G. and Adams, S. E.: Interobserver and intraobserver differences in the diagnosis of urothelial cells. Comparison with classification by computer. Anal Quant Cytol, 6: 112, 1984 12. Baltaci, S., Suzer, O., Ozer, G. et al: The efficacy of urinary cytology in the detection of recurrent bladder tumours. Int Urol Nephrol, 28: 649, 1996 13. Wiener, H. G., Vooijs, G. P. and van’t Hof-Grootenboer, B.: Accuracy of urinary cytology in the diagnosis of primary and recurrent bladder cancer. Acta Cytol, 37: 163, 1993 14. Vogeli, T. A., Grimm, M.-O. and Ackermann, R.: Prospective study for quality control of TUR of bladder tumors by routine second TUR (ReTUR). J Urol, Suppl., 159: 143, abstract 543, 1998 15. Mersdorf, A., Brauers, A., Wolff, J. M. et al: Second TUR for superficial bladder cancer: a must? J Urol, Suppl. 159: 143, abstract 542, 1998 16. Fradet, Y. and Lockhart, C.: Performance characteristics of a new monoclonal antibody test for bladder cancer: ImmunoCyt. Can J Urol, 4: 400, 1997 17. Mian, C., Pycha, A., Wiener, H. et al: Immunocyt: a new tool for detecting transitional cell cancer of the urinary tract. J Urol, 161: 1486, 1999 18. Ellis, W. J., Blumenstein, B. A., Ishak, L. M. et al: Clinical evaluation of the BTA TRAK assay and comparison to voided urine cytology and the Bard BTA test in patients with recurrent bladder tumors. The Multi Center Study Group. Urology, 50: 882, 1997 19. Ishak, I. and Ellis, W. J.: A comparison of the BTA stat and the BTA TRAK assays: two new tests for the detection of recurrent bladder cancer (BC) in urine. J Urol, Suppl., 159: 245, abstract 936, 1998
757
20. Soloway, M. S., Briggman, J. V., Carpinito, G. A. et al: Use of a new tumor marker, urinary NMP22, in the detection of occult or rapidly recurring transitional cell carcinoma of the urinary tract following surgical treatment. J Urol, 156: 363, 1996 21. Landman, J., Kavaler, E., Chang, Y. et al: Sensitivity and specificity of NMP-22, telomerase, and BTA in the detection of human bladder cancer. J Urol, Suppl., 159: 245, abstract 938, 1998 22. Allard, P., Bernard, P., Fradet, Y. et al: The early clinical course of primary Ta and T1 bladder cancer: a proposed prognostic index. Br J Urol, 81: 692, 1998 23. Sobin, L. H. and Wittekind, C.: TNM Classification of Malignant Tumours. 5th ed. International Union Against Cancer New York: Springer-Verlag, 1997 24. Marubini, E.: Analysing survival data from clinical trials and observational studies. New York: J. Wiley, 1995 25. Lockhart, C., Parent, C., Desaulniers, M. et al: Evaluation of a monoclonal antibody test for the detection of superficial bladder cancer: Immunocyt. J Urol, Suppl., 159: 243, abstract 929, 1998 26. Sarosdy, M. F., Hudson, M. A., Ellis, W. J. et al: Improved detection of recurrent bladder cancer using the Bard BTA stat test. Urology, 50: 349, 1997 27. Raitanen, M.-P., Martila, T., Tammela, T. L. J. et al: The Bard BTA stat test in monitoring of bladder cancer. J Urol, Suppl., 159: 244, abstract 935, 1998 28. Shenoy, U. A., Schumann, G. B. and DeBellis, C. C.: Prevalence of hematuria in urothelial neoplasia. Am J Clin Pathol, 85: 80, 1986 29. Messing, E. M. and Vaillancourt, A.: Hematuria screening for bladder cancer. J Occup Med, 32: 838, 1990 30. Vera-Donoso, C. D., Llopis, B., Oliver, F. et al: Follow-up of superficial bladder cancer: how to spare cystoscopies? Eur Urol, 17: 17, 1990 EDITORIAL COMMENT The current strategy for following patients with superficial bladder cancer continues to be based on frequent cystoscopic evaluations using a schedule that is largely empirical. This situation persists because of the documented ability of cystoscopy to detect recurrent disease in asymptomatic individuals and the perceived inability to predict disease recurrence reliably by other methods. The recent development of a variety of sensitive markers for bladder cancer has raised the hope that followup based on individual risk may finally become possible. The authors explored this possibility in patients with superficial bladder cancer who had a low to moderate risk of tumor recurrence (disease-free at first followup cystoscopy). They constructed a computer based model comparing standard followup with cystoscopy indicated by a positive marker test. The data demonstrate the cost-effectiveness of this strategy. This study is a good first step toward developing individualized therapy based on marker analysis and suggests that this strategy is worth pursuing. However, as with many computer models, the mathematics are internally consistent but may not reflect the vagaries of the real world. The risk of disease progression and its impact on morbidity, mortality and expense cannot be reliably assessed in this model. Nevertheless, it is time to take up the gauntlet. Prospective clinical studies using this or similar designs preferably with randomization against conventional therapy are urgently needed to demonstrate the cost-effectiveness and safety of marker based individualized treatment. The data obtained from such a study will be of significant benefit to our patients. H. Barton Grossman Department of Urology University of Texas M. D. Anderson Cancer Center Houston, Texas REPLY BY AUTHORS We agree regarding the need for clinical trials to compare the safety and efficacy of urinary markers and standard followup regimens with cystoscopy and cytology for this patient cohort. A significant hurdle in designing such trials is the lack of data on the natural history of superficial bladder cancer recurrence monitored by urinary markers. Physicians or patients may be unwilling to forego surveillance cystoscopy and cytology when using urinary markers. Despite these factors, we are currently pursuing prospective collaborative studies to assess how urinary markers affect disease progression and quality of life for patients with superficial bladder cancer.
Vol. 163, 752–757, March 2000 Printed in U.S.A.
COMPARISON OF MOLECULAR AND CONVENTIONAL STRATEGIES FOR FOLLOWUP OF SUPERFICIAL BLADDER CANCER USING DECISION ANALYSIS ROBERT K. NAM, DONALD A. REDELMEIER, PHILIPPE E. SPIESS, HEATHER A. SAMPSON, YVES FRADET* AND MICHAEL A. S. JEWETT From the Division of Urology, Princess Margaret Hospital, University Health Network, Department of Medicine, Sunnybrook and Women’s College Health Sciences Centre, University of Toronto, Toronto, Ontario, and Division of Urology, Laval University, Laval, Quebec, Canada
ABSTRACT
Purpose: Patients with superficial bladder cancer require long-term surveillance for recurrence. We compared the cost of cystoscopy and cytology (standard care) to that of urinary markers (modified care) for patients with a history of superficial bladder cancer. Materials and Methods: We constructed a decision analysis model that compared the 2 strategies for a hypothetical followup interval of 3 years. Probabilities required for the decision tree were based on a cohort of 361 patients diagnosed with superficial bladder cancer from 1987 to 1997. Sensitivity analyses were used to determine whether test sensitivity and specificity would affect cost thresholds. Costs for each strategy were then applied to actual practice patterns. Results: The cost of modified care ranged from $158 to $228 for each followup visit when using a urinary marker with a sensitivity and specificity of 95% and 77%, respectively. The cost of standard care was $240 for each followup visit. Based on sensitivity analyses the probability of disease recurrence and urinary marker accuracy were important determinants of expected costs. Mean number of followup assessments for patients followed more than 3 years was 4.3, 2.2 and 1.5 for years 1, 2 and 3, respectively. Cumulative costs of modified care were lower than those of standard care. Conclusions: Urinary marker testing for followup of patients with superficial bladder cancer is less expensive than the standard method of cystoscopy and urinary cytology based on our model. Future studies will be required to consider other factors that could affect the cost advantage of urinary markers, including indirect costs, the psychosocial impact of testing and different surveillance frequencies. KEY WORDS: bladder neoplasms; tumor markers, biological; decision support techniques; costs and cost analysis
Bladder cancer is common, lethal and potentially curable.1– 4 Up to 80% of all superficial bladder cancers will recur and 11% to 28% will progress1– 4 and, thus, early detection and treatment of recurrence are essential.5 This rationale is based on premises that bladder cancer progression results in metastasis and death, and treating disease recurrence delays progression. The former premise has been well established as 78% of patients with progression of superficial bladder cancer eventually die of bladder cancer.1 The latter premise has yet to be directly established to our knowledge. Since there is no untreated natural history data to compare whether treating disease recurrence prevents clinical progression, we can only infer this premise indirectly. Holma¨ng et al showed an inverse relationship between the rate of disease progression, and tumor size and volume among patients treated for superficial bladder cancer.2 Thus, detecting disease recurrence earlier would result in smaller tumor size and volume and, subsequently, lower progression rates. Accordingly, patients undergo surveillance cystoscopy and urinary cytology every 3 months for 1 to 2 years, every 6 months for the subsequent years and every year thereafter.6, 7 Such
testing results in anxiety, discomfort and costs.7 Furthermore, cystoscopy is invasive and can cause urinary tract infection in 15% of patients, dysuria in 59%, urinary frequency in 41% and hematuria in 30%.8, 9 Urinary cytology is an ancillary method for detecting recurrence but is associated with significant inaccuracy due to specimen handling techniques,10 variable observer reliability,11 large test performance differences between tumor stages and grades, and variation induced by intravesical treatments.12, 13 Other strategies to improve surveillance have been proposed. Adjuvant intravesical therapy and re-resection of the initial tumor have been shown to reduce recurrence rates and allow less frequent followup.4, 7, 14, 15 Tailoring the frequency of surveillance cystoscopy according to the risk of recurrence also appears safe.5, 7 More recently urinary markers for disease recurrence have been developed and commercialized16 –20 which require use of immunoassay techniques to detect tumor cells.16, 18, 19, 21 Marker testing only requires voided urine specimens to produce objective results and has the potential to reduce health care costs as well as morbidity associated with invasive surveillance methods. However, urinary marker testing needs to be accurate to serve as a safe and cost-effective adjunct or as a substitute for followup cystoscopy. To understand the use of these markers better we compared the costs of urinary markers and followup cystoscopy to detect disease recurrence in a cohort of patients with
Accepted for publication October 15, 1999. Supported by a grant from Diagnocure, Inc., Sainte-Foy, Quebec, and the Surgical Scientist Program, Department of Surgery, University of Toronto, Canada. * Financial interest and/or other relationship with Diagnocure, Inc. 752
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
superficial bladder cancer through a cost minimization analysis using decision analytic techniques. MATERIALS AND METHODS
Historical cohort. We studied patients treated at the Toronto Hospital and by the Quebec Urologic Group22 between 1983 and 1997 who were disease-free at first followup after initial treatment for superficial transitional carcinoma of the bladder (stages Ta and T123). The primary end point of biopsy proved recurrence was analyzed using nonconstant hazard functions and Kaplan-Meier survival analysis from the time of the first disease-free assessment.24 An event was defined as recurrence. Patients were considered to be at risk from the date of first followup until that of recurrence or last followup. The effect of tumor stage and grade was examined by univariate and multivariate analyses using computer software. Decision model. We constructed a decision analysis model to estimate the costs of followup with cystoscopy and urinary cytology (standard care), and a urinary molecular marker (modified care) to diagnose recurrence (fig. 1). The outcome for the decision tree was disease recurrence. We used the societal perspective, assumed a time horizon of 3 years and applied no discount factor. Entry criteria for the decision tree consisted of disease-free status at first followup visit, pathological stage Ta or T1 transitional cell carcinoma of the bladder with no evidence of carcinoma in situ and a positive test result with the urinary marker at diagnosis (90% detection rate17). The base case analysis was based on a patient of any age with superficial bladder cancer of any grade and stage requiring surveillance for disease recurrence with followup cystoscopy and cytology or the urinary marker used at the Toronto Hospital (Immunocyt†) with a reported 95% sensitivity and 77% specificity.16 The base case describes the patient for whom the expected costs derived from the model will apply. Probabilities for each branch of the decision tree were based on the hazard rates for recurrence in our cohort, and the reported sensitivity and specificity of our urinary marker.16 Various hazard functions were tested and the func† Diagnocure Inc., Quebec, Canada
753
tion that yielded the highest correlation to Kaplan-Meier survival estimates was used. Baseline assumptions. Major assumptions for the decision tree included that the gold standard of cystoscopy and cytology is 100% sensitive and specific; some tumors will test marker negative and may continue to be negative despite recurrence which would lead to false-negative results; a positive result from the urinary marker requires confirmatory cystoscopy; undetected recurrent tumors will not always produce hematuria or symptoms before progression, specifically 20% will be assumed to progress silently; all patients with progression will require radical cystectomy; patients will undergo followup evaluation every 3 months for year 1, every 6 months for year 2 and once for year 3; and if a patient has recurrence during surveillance, the risk after treatment will be the same as the initial risk of recurrence. Standard care consisted of cystoscopy at a hospital ambulatory facility and urinary cytology at each patient visit. The costs comprised professional fees of the urologist and pathologist, personnel time for cystoscopy and pathology, and institutional overhead costs, excluding depreciation. The cost of communicating test results, such as transportation and parking costs to the patient, was not included. Modified care costs included the reported cost ranges of commercially available urinary markers (kit costs assumed the entire kit was used).16, 18 –20, 25 Laboratory staff costs were estimated by the manufacturer and included the per test charge, and all handling and transportation fees. Sensitivity analysis. We examined all key variables of the model over a wide range of values using sensitivity analysis. Plausible ranges of the urinary marker sensitivity and specificity were based on reported ranges of other commercially available urinary markers.16, 18 –20, 26 We determined the cost thresholds (point at which 2 strategies have equal expected costs) for a range of sensitivities and specificities of the urinary marker, which was performed within the covariate categories (disease stage and grade) that were found to be significant prognostic factors for tumor recurrence within our cohort. Because costs of standard and modified care vary between centers, we tested the sensitivity analysis over a range of costs for each management option.
FIG. 1. Decision tree for patients at each followup visit. UM, urinary marker
754
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
TABLE 1. Baseline characteristics among patients with superficial bladder cancer who were disease-free at first followup evaluation Disease Stage Ta T1 Totals
No. Histological Grade (%) 1
2
3
Totals
115 7 122 (34)
175 29 204 (56)
10 25 35 (10)
300 (83) 61 (17) 361
RESULTS
Likelihood for disease recurrence. A total of 361 patients were disease-free at first followup assessment after initial resection and satisfied inclusion into our historical cohort analysis (table 1). Mean patient age at diagnosis was 64.9 years (range 29.3 to 93.2) and mean followup was 17.6 months (range 0.1 to 119). Based on correlational analysis of the survival estimates of various hazard functions and Kaplan-Meier survival estimates, the log logistic hazard model showed the highest correlation (Pearson r ⫽ 0.987, p ⫽ 0.0001). Univariate analysis based on the log logistic hazard model showed tumor stage (p ⫽ 0.0036) (fig. 2) and grade (p ⫽ 0.007) to be significant predictors of recurrence-free survival. Multivariate analysis revealed that tumor stage (p ⫽ 0.05) was a more important predictor than grade (p ⫽ 0.09). Costs of care. At the time of analysis the cost of cystoscopy was $221, including $89 for the professional fee, and $132 for facility and equipment. Urinary cytology was $19 per test, including $7 for technical and $12 for professional fees. Total cost of radical cystectomy was $9,570, which included $1,900 for professional fees and $7,670 for institutional fees (hospital wards, blood bank, laboratory, radiological expenses and so forth). The cost of the urinary marker used was $6016 and others ranged from $12 to $125.20, 27 Baseline analysis and sensitivity analysis. Because tumor stage was a stronger predictor of disease recurrence than grade based on the survival model, the cost of modified care was calculated for each followup visit by tumor stage. For the base case the cost of modified care was highest at the 3-month followup visit ($228 for stage Ta, $237 for T1) and decreased for every subsequent visit (table 2). The expected cost of modified care was consistently lower than that of standard care, which was $240 for every followup visit. Based on sensitivity analysis the probabilities for recurrence and urinary marker sensitivity and specificity were
important determinants of the expected costs of modified care. The later the followup evaluation, the lower the sensitivity and specificity of the urinary marker required for modified care to have a cost savings over standard care (table 3). Using the base case 95% sensitivity and 77% specificity as references for 1-way sensitivity analysis, the minimum urinary sensitivity required for modified care to have a cost advantage was 70% at the 18-month followup and minimum specificity was 37% at the 36-month followup (table 3). To compare other costs of urinary markers and standard care, multiple 2-way sensitivity analyses for urinary marker sensitivity and specificity were performed using a fixed recurrence rate of 45.8%, the highest probability (and most conservative) for all followup evaluations. A larger range of urinary marker sensitivities and specificities were plausible at lower costs of the urinary marker ($20) and higher costs of standard care ($300) (fig. 3). As urinary marker costs increased and standard care costs decreased, the urinary marker sensitivities and specificities required for modified care to have a cost savings increased. To determine how the use of modified care would affect actual practice patterns we compared the cumulative costs of modified and standard care for our cohort. Mean number of followup evaluations (cystoscopy and cytology) for patients followed at least 3 years was 4.27 (range 2 to 6) for year 1, 2.20 (range 1 to 4) for year 2 and 1.48 (range 1 to 4) for year 3. This number was determined individually by examining the frequency of followup for the first 3 years. Using our base case urinary marker the cumulative cost of modified care was less than that of standard care after 6 months of followup (fig. 4). DISCUSSION
Urinary molecular markers may provide cost savings compared to standard cystoscopy and urinary cytology for superficial bladder cancer followup. The first assumption based on clinical perception is that to consider modified care urinary marker sensitivity must be close to 90% or better. The disutility of a false-positive or negative result was not included in our analysis, primarily because we are unaware of quantitative reports of the psychosocial impact of false-positive or negative test results with urinary markers or cystoscopy which makes it difficult to calculate the cost impact. Because these indirect costs were not included in our analysis, we did not consider costs to patients for time spent undergoing cytoscopy, including lost wages, transportation and so forth. Also, it is noteworthy
FIG. 2. Kaplan-Meier recurrence-free survival by stage of superficial bladder cancer
755
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP TABLE 2. Base case analysis of expected cost of modified care using the Immunocyt urinary marker Followup Mo.
Stage Ta Ca ($)
Stage T1 Ca ($)
3 228 6 203 9 185 12 173 18 165 24 158 36 158 Expected cost for standard care was $240 for all followup
237 207 188 175 166 159 159 visits.
that our model does not consider how the use of urinary markers might affect disease specific survival. We assumed that the risks of progression and death were equal with both strategies and to consider how this would affect our model, an analysis of quality adjusted life expectancy would be required, which is beyond the scope of our study. Given these limitations, we have provided a simple method for decision markers to evaluate whether urinary markers would have cost savings over standard care for an institution. The acceptable cost of a marker with known test performance
TABLE 3. One-way sensitivity analyses of urinary marker sensitivities and specificities for each followup visit Followup Mo. 3
6
9
12
18
24
36
% Stage Ta sensitivity 93 88 82 76 70 Not significant Not significant % Stage Ta specificity 69 54 46 41 38 36 36 % Stage T1 sensitivity 95 89 83 77 71 Not significant Not significant % Stage T1 specificity 75 56 47 42 39 37 37 Thresholds were defined as the minimal sensitivity and specificity for which the expected cost of modified care is less than that of standard care, and baseline was 95% sensitivity and 77% specificity.
FIG. 3. Comparison of cost of cystoscopy and cytology (SC, $100 to $300) to that of urinary marker (MC, $20 to $100) using multiple 2-way sensitivity analyses. Cost advantage of bladder cancer followup by urinary marker is represented by black areas according to urinary marker accuracy. Probability of disease recurrence is fixed at 46%.
756
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
FIG. 4. Cumulative cost comparison between standard care (SC) and modified care (MC) using costs at Toronto Hospital, urinary marker test performance of 95% sensitivity and 77% specificity, and actual practice patterns of followup frequency.
can be calculated using local costs of standard care (cystoscopy plus cytology, fig. 3). Alternatively, with known local costs of standard care and a urinary marker, the required minimum test performance for that marker can be calculated. The use of other commercially available urinary markers appears to be financially viable depending on institutional costs of standard and modified care. Other assumptions made for the decision tree have been biased in favor of standard care and several need to be examined. In our model no confirmatory cystoscopy or cytology is done for patients with a negative marker test. Thus, tumor recurrence might be missed which could subsequently progress. Nearly all undetected tumors will ultimately present with hematuria or other symptoms, and the overall progression rate of low risk stage Ta, T1, grade 1–2 tumors is exceedingly low.28, 29 Furthermore, in a study that compared surveillance cystoscopy and a urinary lipid marker for disease recurrence among patients with superficial bladder cancer for 6 months after initial treatment the 27% with a false-negative test result had no clinical or stage progression.30 We assumed a 20% progression rate for patients with a false-negative urinary marker result and that they were curable with radical cystectomy. Based on available published data the 20% rate is probably a significant overestimate.22 Although the risk of recurrence positively correlated to the number of previous recurrences,3 we did not stratify each pretest likelihood of recurrence by the number of recurrences, because a variety of patient and physician dependent factors made it difficult. Patients may have missed scheduled appointments due to illness or vacation, or physicians may be overly cautious and choose to do more frequent followup evaluations. To assign and rank each visit would require an iterative process of examining each patient pattern of followup that would be difficult in a large cohort. These factors are also likely to explain why the frequency of followup evaluations was higher than expected. Furthermore, we did not stratify patients based on other important prognostic factors, such as tumor grade,3 because of the limited sample size of our cohort. Thus, we used the strongest prognostic factor (tumor stage) found from our cohort for our decision model. It would be important for future studies with a larger sample size to consider other important prognostic factors.
Also, the probability of recurrence was the most important determinant of cost associated with modified care using urinary markers. In our study the use of urinary markers was applied to the entire cohort. However, the cost of urinary markers would be significantly reduced if applied to the 60% of patients at low risk of recurrence based on simple clinical findings, such as the number of tumors and results of the first followup cystoscopy.3 Furthermore, for patients with low grade papillary tumors cytology may not provide additional diagnostic information given its low sensitivity, and cystoscopy alone may be used which could affect the cost advantage of urinary markers.12 However, because the most important predictors of cost within the decision tree were the probability of disease recurrence, and urinary marker sensitivity and specificity, the reduction in cost of standard care for cystoscopy alone did not significantly affect expected costs. As well, the gold standard of cystoscopy and cytology could be used at different followup intervals. Olsen et al showed that lowering the frequency of followup cystoscopy in select patients may not affect disease progression rates.5 It would be important to consider these possibilities further in future studies. Finally, molecular techniques may even provide a superior method for early detection of recurrence. Although we used the combination of cystoscopy and cytology as the gold standard, microscopic lesions may still be missed. It is possible that a urinary marker test may detect the presence of occult bladder cancer recurrence when conventional methods do not. In such a case a true positive result with a urinary marker test may be classified incorrectly as false-positive if the gold standard cystoscopy is negative. Early reports on urinary marker followup of patients after initial resection have revealed that some patients with negative cystoscopy but a positive marker urinary test have bladder lesions with further followup.26 In our analysis this bias would favor the more conservative standard care method. CONCLUSIONS
From the payer perspective urinary marker testing for followup of patients with superficial bladder cancer can provide lower costs than the standard method of cystoscopy and urinary cytology. Reported sensitivities and specificities of various uri-
MOLECULAR AND CONVENTIONAL STRATEGIES FOR BLADDER CANCER FOLLOWUP
nary markers are sufficient to be cost saving per patient within 3 years of followup. The cost advantage of urinary markers over standard care remains over a wide range of service costs and increases with time of followup. The actual frequency of followup for patients on surveillance for superficial bladder cancer may be higher than what is recommended. In this setting urinary marker testing showed an even higher cumulative cost savings over standard care. This method is also more appealing to patients since only a voided urine specimen is required, and it avoids the morbidity associated with invasive testing. REFERENCES
1. Herr, H. W.: Natural history of superficial bladder tumors: 10- to 20-year follow-up of treated patients. World J Urol, 15: 84, 1997 2. Holma¨ng, S., Hedelin, H., Anderstro¨m, C. et al: The relationship among multiple recurrences, progression and prognosis of patients with stages TA and T1 transitional cell cancer of the bladder followed for at least 20 years. J Urol, 153: 1823, 1995 3. Kurth, K. H., Denis, L., Bouffioux, C. et al: Factors affecting recurrence and progression in superficial bladder tumours. Eur J Cancer, 31A: 1840, 1995 4. Pawinksi, A., Sylvester, R., Kurth, K. H. et al: A combined analysis of European Organization for Research and Treatment of Cancer, and Medical Research Council randomized clinical trials for the prophylactic treatment of stage TaT1 bladder cancer. J Urol, 156: 1934, 1996 5. Olsen, L. H. and Genster, H. G.: Prolonging follow-up intervals for non-invasive bladder tumors: a randomized controlled trial. Scand J Urol Nephrol, Suppl., 172: 33, 1995 6. Messing, E. M. and Catalona, W.: Urothelial tumors of the urinary tract. In: Campbells Urology. Edited by P. C. Walsh, A. B. Retik, E. D. Vaughan, Jr. et al. Philadelphia: W. B. Saunders, vol. 3, chapt. 77 pp. 2327–2410. 1998 7. Hall, R. R., Parmar, M. K., Richards, A. B. et al: Proposal for changes in cystoscopic follow up of patients with bladder cancer and adjuvant intravesical chemotherapy. BMJ, 308: 257, 1994 8. Westenfelder, M., Rosset, K. and Pelz, K.: Development of nosocomial and iatrogenic urinary tract infections (UTI) following urological interventions. A prospective clinical study. Scand J Urol Nephrol, Suppl., 104: 59, 1987 9. Denholm, S. W., Conn, I. G., Newsam, J. E. et al: Morbidity following cystoscopy: comparison of flexible and rigid techniques. Br J Urol, 66: 152, 1990 10. Leyh, H. and Mazeman, E.: Bard BTA test compared with voided urine cytology in the diagnosis of recurrent bladder cancer. Eur Urol, 32: 425, 1997 11. Sherman, A. B., Koss, L. G. and Adams, S. E.: Interobserver and intraobserver differences in the diagnosis of urothelial cells. Comparison with classification by computer. Anal Quant Cytol, 6: 112, 1984 12. Baltaci, S., Suzer, O., Ozer, G. et al: The efficacy of urinary cytology in the detection of recurrent bladder tumours. Int Urol Nephrol, 28: 649, 1996 13. Wiener, H. G., Vooijs, G. P. and van’t Hof-Grootenboer, B.: Accuracy of urinary cytology in the diagnosis of primary and recurrent bladder cancer. Acta Cytol, 37: 163, 1993 14. Vogeli, T. A., Grimm, M.-O. and Ackermann, R.: Prospective study for quality control of TUR of bladder tumors by routine second TUR (ReTUR). J Urol, Suppl., 159: 143, abstract 543, 1998 15. Mersdorf, A., Brauers, A., Wolff, J. M. et al: Second TUR for superficial bladder cancer: a must? J Urol, Suppl. 159: 143, abstract 542, 1998 16. Fradet, Y. and Lockhart, C.: Performance characteristics of a new monoclonal antibody test for bladder cancer: ImmunoCyt. Can J Urol, 4: 400, 1997 17. Mian, C., Pycha, A., Wiener, H. et al: Immunocyt: a new tool for detecting transitional cell cancer of the urinary tract. J Urol, 161: 1486, 1999 18. Ellis, W. J., Blumenstein, B. A., Ishak, L. M. et al: Clinical evaluation of the BTA TRAK assay and comparison to voided urine cytology and the Bard BTA test in patients with recurrent bladder tumors. The Multi Center Study Group. Urology, 50: 882, 1997 19. Ishak, I. and Ellis, W. J.: A comparison of the BTA stat and the BTA TRAK assays: two new tests for the detection of recurrent bladder cancer (BC) in urine. J Urol, Suppl., 159: 245, abstract 936, 1998
757
20. Soloway, M. S., Briggman, J. V., Carpinito, G. A. et al: Use of a new tumor marker, urinary NMP22, in the detection of occult or rapidly recurring transitional cell carcinoma of the urinary tract following surgical treatment. J Urol, 156: 363, 1996 21. Landman, J., Kavaler, E., Chang, Y. et al: Sensitivity and specificity of NMP-22, telomerase, and BTA in the detection of human bladder cancer. J Urol, Suppl., 159: 245, abstract 938, 1998 22. Allard, P., Bernard, P., Fradet, Y. et al: The early clinical course of primary Ta and T1 bladder cancer: a proposed prognostic index. Br J Urol, 81: 692, 1998 23. Sobin, L. H. and Wittekind, C.: TNM Classification of Malignant Tumours. 5th ed. International Union Against Cancer New York: Springer-Verlag, 1997 24. Marubini, E.: Analysing survival data from clinical trials and observational studies. New York: J. Wiley, 1995 25. Lockhart, C., Parent, C., Desaulniers, M. et al: Evaluation of a monoclonal antibody test for the detection of superficial bladder cancer: Immunocyt. J Urol, Suppl., 159: 243, abstract 929, 1998 26. Sarosdy, M. F., Hudson, M. A., Ellis, W. J. et al: Improved detection of recurrent bladder cancer using the Bard BTA stat test. Urology, 50: 349, 1997 27. Raitanen, M.-P., Martila, T., Tammela, T. L. J. et al: The Bard BTA stat test in monitoring of bladder cancer. J Urol, Suppl., 159: 244, abstract 935, 1998 28. Shenoy, U. A., Schumann, G. B. and DeBellis, C. C.: Prevalence of hematuria in urothelial neoplasia. Am J Clin Pathol, 85: 80, 1986 29. Messing, E. M. and Vaillancourt, A.: Hematuria screening for bladder cancer. J Occup Med, 32: 838, 1990 30. Vera-Donoso, C. D., Llopis, B., Oliver, F. et al: Follow-up of superficial bladder cancer: how to spare cystoscopies? Eur Urol, 17: 17, 1990 EDITORIAL COMMENT The current strategy for following patients with superficial bladder cancer continues to be based on frequent cystoscopic evaluations using a schedule that is largely empirical. This situation persists because of the documented ability of cystoscopy to detect recurrent disease in asymptomatic individuals and the perceived inability to predict disease recurrence reliably by other methods. The recent development of a variety of sensitive markers for bladder cancer has raised the hope that followup based on individual risk may finally become possible. The authors explored this possibility in patients with superficial bladder cancer who had a low to moderate risk of tumor recurrence (disease-free at first followup cystoscopy). They constructed a computer based model comparing standard followup with cystoscopy indicated by a positive marker test. The data demonstrate the cost-effectiveness of this strategy. This study is a good first step toward developing individualized therapy based on marker analysis and suggests that this strategy is worth pursuing. However, as with many computer models, the mathematics are internally consistent but may not reflect the vagaries of the real world. The risk of disease progression and its impact on morbidity, mortality and expense cannot be reliably assessed in this model. Nevertheless, it is time to take up the gauntlet. Prospective clinical studies using this or similar designs preferably with randomization against conventional therapy are urgently needed to demonstrate the cost-effectiveness and safety of marker based individualized treatment. The data obtained from such a study will be of significant benefit to our patients. H. Barton Grossman Department of Urology University of Texas M. D. Anderson Cancer Center Houston, Texas REPLY BY AUTHORS We agree regarding the need for clinical trials to compare the safety and efficacy of urinary markers and standard followup regimens with cystoscopy and cytology for this patient cohort. A significant hurdle in designing such trials is the lack of data on the natural history of superficial bladder cancer recurrence monitored by urinary markers. Physicians or patients may be unwilling to forego surveillance cystoscopy and cytology when using urinary markers. Despite these factors, we are currently pursuing prospective collaborative studies to assess how urinary markers affect disease progression and quality of life for patients with superficial bladder cancer.